KR100485594B1 - A method for removing noise in image and a system thereof - Google Patents

Abstract

The present invention relates to a noise processing method and a system for removing noise in an image, and more particularly, to remove the noise without damaging the edge or detail of the image to improve the sharpness of the image. The present invention relates to a sustainable noise processing method and a system thereof.

A noise processing method for removing noise in an image according to the present invention comprises the steps of: dividing image data included in the image into a predetermined unit area; Calculating thresholds corresponding to the unit region by using pixel values of pixels included in the unit region; Detecting whether a first pixel including an impulse noise exists in the unit region by using the calculated threshold value; Applying a median filter to the first pixel when the first pixel is detected; And identifying a second pixel adjacent to the first pixel; And applying a mean-variance filter to the second pixel.

The present invention provides a noise processing method and a system for performing motion detection by efficiently removing Poisson noise while preserving an image in a low brightness region.

Description

Noise processing method and system for removing noise in image {A METHOD FOR REMOVING NOISE IN IMAGE AND A SYSTEM THEREOF}

The present invention relates to a noise processing method and system for removing noise in an image, and more particularly, to a noise processing method capable of maintaining image clarity by removing noise without damaging edges or detailed information, and a system thereof. It's about the system.

Today, with the development of video technology, various video media such as high performance camera, digital camera, CCTV, video capture system, etc. are being developed. The most important problem of such image capturing media is the accurate reconstruction of the image of the photographed subject. In particular, noise that may be added to an image not only degrades the quality of the image, but also reduces the compression efficiency. Therefore, removing noise quickly and accurately is the most important problem for restoring the image.

However, when spatial filtering or temporal filtering is applied in the process of removing the noise included in the image, the noise is removed but smearing or blur occurs, and thus the sharpness of the image itself is damaged. In particular, this problem was inevitably more pronounced in an image in which a lot of noise was generated in a poor environment where there is not enough light.

It is an object of the present invention to provide a noise processing method and system capable of maintaining sharpness by preserving edges or detailed information of an image while effectively removing impulse noise or Poisson noise in a process of performing spatial filtering on the image. .

In order to achieve the above object and solve the problems of the prior art, the present invention comprises the steps of dividing the image data included in the image into a predetermined unit area; Calculating thresholds corresponding to the unit region by using pixel values of pixels included in the unit region; Detecting whether a first pixel including an impulse noise exists in the unit region by using the calculated threshold value; Applying a median filter to the first pixel when the first pixel is detected; Identifying a second pixel adjacent to the first pixel; And applying a mean-variance filter to the second pixel.

In addition, the present invention is a noise processing system for removing noise in an image, the image data included in the image is divided into a predetermined unit region, and the unit by using the pixel value of the pixels included in the unit region A threshold calculator configured to calculate thresholds corresponding to the regions, respectively; A first pixel detecting whether a first pixel including an impulse noise exists in the unit region by using the calculated threshold value, and applying a median filter to the first pixel when the first pixel is detected Filter unit; And a second filter unit identifying a second pixel adjacent to the first pixel and applying a mean-variance filter to the second pixel.

Hereinafter, with reference to the accompanying drawings will be described an embodiment of the present invention;

1 is a block diagram showing the configuration of a noise processing system according to an embodiment of the present invention. Referring to FIG. 1, the noise processing system according to the present invention may include a threshold calculator 101, a first filter unit 102, and a second filter unit 103.

The threshold calculator 101 classifies the image data included in the image into a predetermined unit area, and calculates threshold values corresponding to the unit area by using pixel values of pixels included in the unit area.

The first filter unit 102 detects whether a first pixel including an impulse noise exists in the unit region by using the calculated threshold value, and, when the first pixel is detected, a median filter ( median filter).

The second filter unit 103 identifies a second pixel adjacent to the first pixel, and applies a mean-variance filter to the second pixel.

Hereinafter, the operation of each component of the noise processing system will be described in more detail with reference to FIG. 2. 2 is a flowchart illustrating a noise processing process by the noise processing system according to the present invention.

In operation S201, the threshold calculator 101 of the noise processing system divides image data included in an image into a predetermined unit area.

In operation S202, the threshold calculator 101 calculates a threshold value of the unit area by using pixel values of pixels in each unit area. The threshold is used to determine whether there is an impulse noise in the unit region. The threshold may be calculated as in Equation 1.

remind Is a minimum threshold for sensing a pixel containing impulse noise, and Is an average value of pixel values of each pixel included in the unit region. As such, since the threshold is dynamically calculated for each unit area, the noise processing system according to the present embodiment can more accurately detect the impulse noise.

The first filter unit 102 determines that the pixel value of the predetermined pixel included in the predetermined unit region has a value higher than the threshold calculated by Equation 1 for the unit region as a pixel in which an impulse noise is generated. If the value is lower than the threshold value, it is determined as a pixel in which no impulse noise occurs (S203). Hereinafter, the pixel in which the impulse noise occurs is referred to as a "first pixel".

When the pixel is detected as the first pixel, the first filter unit 102 removes impulse noise by applying a median filter to the first pixel to replace the value of the first pixel with a median value (S204). . The first filter unit 102 may use a median filter such as Equation 2.

In Equation 2, W denotes a unit area, and (i, j) denotes coordinates of the first pixel in the entire image. Y means the changed value of the first pixel, that is, the impulse noise can be removed by changing the first pixel value to Y.

Once the first pixel is detected, the second filter unit 103 applies an average-variance filter to the peripheral pixels of the first pixel. That is, the first filter unit 102 and the second filter unit 103 are interlocked.

In general, since the Poisson noise is often generated in the peripheral pixels of the pixel in which the impulse noise occurs, in the present embodiment, the first filter unit 102 averages the peripheral pixels of the pixel in which the impulse noise occurs. -Apply a variance filter to remove Poisson noise at once. In order to remove the Poisson noise, a method of approximating the Poisson noise distribution to a Gaussian distribution having a mean and a variance may be used.

The second filter unit 103 detects the second pixels, which are neighboring pixels adjacent to the first pixel, in step S205, and applies an average-variance filter to the second pixels in step S206, thereby including the impulse noise. The values of the neighboring pixels (LNDs) are corrected based on. For example, an average-dispersion filter may be applied to peripheral pixels, that is, eight pixels, in a region having a size of three pixels by three pixels around the first pixel.

The second filter unit 103 may use an average-variance filter as shown in Equation 3 below.

Wow Is the coordinate of the first pixel Represents the variance and mean of the second pixels of a region of a predetermined predetermined size of.

In an area of a certain size without change Is calculated, and Equation 3 can be expressed as follows.

Equation 4 means that the pixel value of the surrounding pixel is close to the average value in that region, and Poisson noise is effectively averaged (ie, removed).

On the contrary, in the area with edge or detail in the image data, Since E has a large value, Equation 3 can be approximated as follows.

As can be seen from Equation 5, since the second pixel value is maintained in the area including the edge or the detailed information, that is, the region with many changes, the edge or the detailed information are preserved.

With the above configuration, the noise processing system according to the present invention applies a median filter to the first pixel including the impulse noise to remove the impulse noise, and averages the second pixels as the peripheral pixels of the first pixel. Apply a variance filter to remove Poisson noise.

Since the second filter unit 103 applies the average-variance filter to the second pixel around the first pixel only when the first pixel is sensed, the median filter for removing the impulse noise according to the present invention is consequently according to the present invention. The average-variance filter for removing the and Poisson noises is working together.

In particular, the average-dispersion filter according to the present invention can preserve the edge or the detailed information in that the second pixel around the first pixel is replaced with the average value only in an area where no edge or detailed information is present.

3 is a diagram illustrating noise processing results of a unit region including edge and detailed information and a unit region including no edge and detailed information according to an exemplary embodiment of the present invention.

Referring to FIG. 3, the noise removing process of the unit region 310 including the edge and the detail information and the unit region 320 including the edge and the detail information are performed differently so that the noise processing system removes the Poisson noise. While protecting the detailed information and the edge of the image, it will be described in detail below. 3 (a) shows the original unit region 310, and (b) shows the unit region 310 after the median filter and the mean-dispersion filter are applied. 3C shows the original unit region 320 and (d) shows the unit region 320 after the median filter and the mean-dispersion filter are applied.

First, a noise removing process of the unit region 310 that does not include edge and detailed information of the input image will be described. That is, the case where the first pixel 301 including the impulse noise is generated in the unit region 310 including no edge and detailed information will be described as an example. The first filter unit 102 of the noise processing system identifies the first pixel 301 of the unit region 310 by using a threshold calculated as in Equation 1, and compares the value of the first pixel 301 with the above. Correct the median value using the same median filter.

The second filter unit 103 replaces the pixel value N of the second pixel 302 with an average value M by applying an average-dispersion filter to the second pixels 302 that are adjacent pixels of the first pixel. . That is, since the variation of each pixel value of the second pixels 302 is small in the unit area 310 in which no edge or detail information exists, the dispersion value of the second pixels is reduced, so that the second pixel as shown in Equation 4 This value N is corrected to an average value M of the second pixels.

As a result, the impulse noise and the Poisson noise included in the unit region 310 are effectively removed.

Secondly, a process of removing noise in the unit region 320 including edge and detailed information of the input image will be described. The first pixel 303 including the impulse noise of the unit region 320 is corrected to the median value by the median filter of the first filter unit 102 as described above. The second pixel 304, which is an adjacent pixel of the first pixel 303 of the unit region 320, is also corrected by an average-dispersion filter of the second filter unit 103, wherein the second pixel of the unit region 320 is adjusted. Since the pixel 304 belongs to a region where the change is large due to the edge or the detailed information, the dispersion value of the second pixels becomes large, so that the corrected value as shown in Equation 5 maintains the value N of the original second pixel. As a result, in the region where the image characteristics such as the edge and detail information are large, the original pixel value is maintained to remove the impulse noise while maintaining the detail and edge sharpness.

In addition, hereinafter, an image processing system using a noise processing system according to the present invention will be described. In particular, the image processing system according to the present embodiment may be photographed in a low light environment to effectively provide impulse noise or Poisson noise from an image in which the image source itself is not clean.

FIG. 4 is a diagram illustrating a case where an image processing system is used as a configuration of a prefilter. The image captured by the camera 401 is captured by the capture module 402, and the image processing system 403 including the noise processing system according to the present invention is processed before the image is encoded by the video encoder, such as being digitized or compressed. In order to obtain a good image and to improve compression efficiency, preprocessing is performed such as removing noise from the image.

The image encoded by the video encoder 404 may be stored in a storage space of the system or may be transmitted to a remote system via a network. An image stored in the storage space, an image transmitted to the remote system, or an image transmitted and stored to the remote system may be decoded by the video decoder 405 and displayed on a predetermined display means 406. Meanwhile, FIG. 4 illustrates a case in which the image processing system 403 according to the present invention is used as a configuration of a preprocessing filter, but this is merely an example, and the present invention effectively removes noise from an image captured by a photographing apparatus. It can be widely used to obtain images that can increase the sharpness.

Hereinafter, operations of the image processing systems 403 and 500 according to the present embodiment will be described in more detail with reference to FIG. 5. 5 is a block diagram illustrating an image processing system 500 according to an exemplary embodiment.

The image processing system 500 includes a segmentation filter 501, a noise processing system 502, a motion detector 503, and a second filter 504.

The segmentation filter 501 divides an image sequentially inputted for each frame into segments composed of a predetermined number of pixels, and divides each segment into a low light region or a high intensity using brightness of image data included in each segment. It is divided into a high light region. Contrast, brightness, or focus measure values of each pixel may be used to determine brightness. In the present specification, "image data of an area including segments divided into low brightness regions" may be briefly expressed as "image data", and also clearly distinguished from image data belonging to a low brightness region or a high brightness region, respectively. For example, an image of an area including both a low brightness area and a high brightness area, that is, an image input to the noise processing system may be expressed as a "full picture". Therefore, the image data may be part or all of the entire image.

The noise processing system 502 removes impulsive noise from image data included in one or more segments divided into low brightness regions.

The motion detector 503 detects a motion fixel among the image data.

The second filter 504 detects false color noise and removes the detected false color noise from the image data.

Hereinafter, the operation of each component of the image processing system 500 as illustrated in FIG. 5 will be described in more detail with reference to FIG. 6.

When the image is sequentially input for each frame, the segmentation filter 501 divides the entire image into segments composed of a predetermined number of pixels (S601). For example, an image captured by the camera 401 may be input to the image processing system 500 at a frame rate of 15 [frame / sec].

Whether the segmentation filter 501 divides the entire image into segments having a size may be selected in consideration of processing efficiency and accuracy, and may be divided into segments having a size of 4 pixels by 4 pixels, for example. .

The segmentation filter 501 recognizes the brightness of the image data included in each segment and moves to the high light region when the brightness exceeds the threshold, and to the low light region when the brightness is below the threshold. (S602).

FIG. 7 is a diagram illustrating an entire image divided into a high brightness region and a low brightness region by the split filter 501. Reference numeral 701 denotes an entire image, and reference numeral 702 denotes an image divided into a high brightness region and a low brightness region. In reference numeral 702, a white portion is a low brightness region and a black portion is a high brightness region. , Is reversed.

In this case, the image processing system 500 skips image processing on image data included in a segment divided into a high brightness region and performs image processing such as noise reduction on only image data included in a segment divided into a low brightness region. The computational efficiency of the entire image processing can be improved. That is, according to the present invention, since the image area to be processed by the image processing system 500 is reduced, the noise is sufficiently removed by processing only image data of a dark area, which is a noisy area, while reducing the time required for calculation. You can also ensure that. In particular, according to the present exemplary embodiment, the image processing of the high brightness region may be omitted by preventing the image processing of the high quality region from being blurred due to the application of the noise filter.

The noise processing system 502 divides the image data included in the segment divided into the low brightness region in the split filter 501 into one or more unit regions and generates pixels including impulse noise in each unit region, that is, the impulse noise is generated. It is detected whether a corroded pixel exists (S603). In this specification, a pixel in which an impulse noise occurs is referred to as a "first pixel".

8 illustrates impulse noise or Poisson noise as described later. Reference numeral 801 in FIG. 8A denotes a first pixel in which impulse noise occurs. FIG. 8B is a diagram illustrating pixel values of pixels belonging to a line including the first pixel 801. Reference numeral 802 denotes that the pixel value in the first pixel has an abnormally large value, that is, an impulse noise has occurred.

The noise processing system 502 senses a first pixel in a predetermined unit area by using a threshold value as shown in Equation 1 below.

As described above, when the pixel is detected as the first pixel, the noise processing system 502 applies a median filter such as Equation 2 to the first pixel to replace the value of the first pixel with the median value. This eliminates impulse noise.

In Equation 2, W denotes a unit area, and (i, j) denotes coordinates of the first pixel in the entire image. Y means the changed value of the first pixel, that is, the impulse noise has been removed.

However, as shown in FIG. 8A, Poisson noise often occurs in the peripheral pixels of the pixel in which the impulse noise occurs. As shown by reference numeral 804 in FIG. 8B, Poisson noise occurs with a distribution similar to the value located at the tail of the Gaussian distribution.

As described above, the noise processing system 502 senses second pixels adjacent to the first pixel to remove such Poisson noise, and applies an average-variance filter such as Equation 3 to the second pixels to impulse them. The pixel values of the second pixels adjacent to the first pixel including the noise are corrected (S605).

With the above configuration, the noise processing system 502 including the noise processing system according to the present invention applies a median filter to the first pixel including the impulse noise to remove the impulse noise, and The Poisson noise is removed by applying an average-variance filter to the second pixels, which are peripheral pixels.

Since the noise processing system 502 applies the mean-variance filter to the second pixel around the first pixel only when the first pixel is detected, the result is a median filter for removing impulse noise according to the present invention. Average-variance filters for removing Poisson noise work together.

The motion detector 503 detects the motion pixel by comparing the image data continuously input in units of frames for each frame (S606). In the present specification, when image data corresponding to an arbitrary frame is referred to as "first image data", the image data corresponding to the frame input immediately after the frame is determined as "second image data".

Detecting a moving pixel (S606) is performed before removing false color noise in the image data. Since filtering to remove false color noise may cause motion blur, it is preferable that false color noise be applied only to a static region in which no moving pixel exists.

In the case of false color noise, unlike impulse noise or Gaussian noise, it is difficult to detect the pixel value of the corresponding pixel or neighboring pixels in the image of one frame, and thus, each frame is compared and detected. Therefore, the previous frame of the current frame must be stored in a predetermined memory means for a certain period or more in order of time, that is, to compare successive frames with each other.

Therefore, when the pixel values of pixels corresponding to each other in a continuously input image are changed, it may be regarded as a normal moving pixel or an undesirable false color noise, and once the moving pixel is detected, the rest is treated as false color noise. can do.

The motion detector 503 may detect a motion pixel by comparing pixels corresponding to each other in a predetermined unit region in a continuously input frame (S607). The unit region for motion detection may be set differently from the unit region used to remove the impulse noise or the Poisson noise and the size thereof. In addition, since the motion detector 503 compares the image of the previous frame with the image of the current frame, the motion detector 503 should temporarily store the image corresponding to the previous frame.

The motion detector 503 according to the present exemplary embodiment may compare pixels in the same unit area on successive different frames by using a threshold value as shown in Equation (6).

SAD (sum absolute difference) represents a threshold determined for each unit area (blocks (m, n)) for motion detection. x ( k, l, t- 1) represents the pixel value of the pixel in the previous frame, and x ( k, l, t ) represents the pixel value of the pixel corresponding to the pixel in the current frame. In addition, A _mn represents a unit region to which the image data of the frame belongs. Therefore, according to equation (6 ) , pixels having coordinates (k, l) are included in the unit region.

As described above, when a pixel, which is a moving pixel, is determined among the pixels whose pixel value is changed between successive frames, the second filter 504 may be configured for a static region that does not include the moving pixel. Temporal filtering can be applied to remove false color noise.

However, as illustrated in FIG. 9, the motion detector 503 may sometimes misjudge a pixel in which false color noise occurs as a motion pixel. In this specification, a pixel falsely determined as a motion pixel as a pixel in which false color noise has been generated is referred to as a "false motion pixel".

9 (a) and 9 (b) show regions corresponding to each other on consecutive frames. The motion detection unit 503, which sequentially recognizes FIGS. 9A and 9B, incorrectly determines that the pixels indicated by the reference numerals 901 and 902 are the pixels where the movement occurs even though the pixels indicated by the false color noise are generated. can do.

In an image of consecutive frames, false color pixels (including false detection motion pixels) should be separated from the motion pixels and removed. When using a low pass filter as a temporal filter as in the prior art, a problem may occur in which smoothing may be excessive and detailed information may not be properly represented. Also, the low pass filter may not be regarded as a false color noise. There is a problem in that the detection motion pixel cannot be distinguished from the normal motion pixel and the false detection motion pixel cannot be removed.

The second filter 504 according to the present embodiment dynamically determines a threshold for each frame as follows, thereby correcting false color pixels including false detection motion pixels and preserving image characteristics such as edges and details in a dark area. do.

Hereinafter, a configuration in which the second filter 504 removes false color noise will be described in detail. For convenience of description, the "first image data" corresponding to a predetermined frame and the image data corresponding to a frame input immediately after the frame are referred to as "second image data".

The second filter 504 converts the first image data to gray scale to generate first gray scale image data, and converts the second image data to gray scale to generate second gray scale image data.

The second filter 504 calculates an overall average value that is an average value of each pixel value of the first gray scale image data. In addition, the second filter 504 calculates, for each unit area, a unit area average value, which is an average value of pixel values of respective pixels of the first gray scale image data. The size of the unit region used to remove the false color noise may be set differently from the unit region used to remove the impulse noise or Poisson noise, or the unit region used for the motion detection. .

When the average value of the unit region exceeds the overall average value, the second filter 504 may include a gray scale among pixels corresponding to the first gray scale image data and the second gray scale image data, that is, a pixel having the same coordinates. Select a pixel with a large value.

When the average value of the unit region is less than or equal to the overall average value, the second filter 504 may include a gray scale value among pixels corresponding to the first gray scale image data and the second gray scale image data, that is, a pixel having the same coordinates. This small pixel is selected.

The second filter 504 replaces the corresponding pixel of the first image data, that is, the pixel in the current frame with the selected pixel.

The above process may be expressed as Equation 7.

x (i, j, t) and y (i, j, t) represent gray scale values of pixels whose coordinates correspond to the first image data and the second image data. Unit area mean, Represents the total mean value. The second filter 504 calculates the unit area average value and the total average value for each frame, and selects a larger value or a smaller value from x ( k, j, t ) and x ( i, j, t- 1). It is determined whether or not, and this configuration is expressed as "dynamically calculate the threshold for each frame."

In this manner, the second filter 504 checks for false color noise (S608) and removes false color noise (S609). In addition, the second filter 504 calculates a threshold value for each frame, thereby effectively detecting and removing false detection motion pixels that were not detected by the prior art.

FIG. 10 is a diagram schematically illustrating a process of removing impulse noise, Poisson noise, and false color noise in image data included in one or more segments divided into low brightness regions.

Referring to FIG. 10, impulse noise and Poisson noise are removed by the noise processing system, and false detection motion pixels mistaken as motion pixels by the motion detector are removed by a second filter using a dynamic threshold. In addition, the noise processing system and the second filter according to the present embodiment are designed to remove noise while preserving the detailed information and the edges, respectively, as described above. As a result, the image processing system 500 according to the present embodiment is used. In this case, it is possible to output a clear image without noise from an image photographed in an environment where there is not enough light.

11 is a view comparing the image processing result of the image processing method and system according to the present invention with the image processing result of the prior art. As described above, according to the present embodiment, after the noise processing system (functioning of the spatial filter) is applied to the image input from the camera, motion detection is performed before the second filter (functioning as the temporal filter) is applied. Has a processing sequence.

Referring to FIG. 11, as a result of processing the image 1101 input from the camera according to the processing sequence as described above, the image 1104 applies a spatial-temporal filter and performs motion detection according to the prior art. As a result of applying the motion detection and the spatiotemporal filter according to the conventional image 1102 or another conventional technique, the noise is removed more effectively than the image 1103 and the image characteristics are well highlighted in the low brightness region. .

As described above, the present invention adopts a method in which a noise processing system that performs a function of a spatial filter is first applied to an image, and motion detection is first performed before applying a second filter, which functions as a temporal filter, to an image. As a result, smearing with respect to the surrounding pixels of the pixel in which the false color noise is generated, which may occur as a result of the application of the second filter, is prevented from affecting the motion detection. Therefore, the motion detection unit can perform more accurate motion detection, and as a result, it is possible to output a quality image as a whole.

12 is a view comparing the relationship between the compression ratio of the image file and the PSNR (Peak Signal to Noise Ratio) of the image processing method and system according to an embodiment of the present invention with the image processing result according to the prior art. . Compared to the prior art, the image processing system according to the present embodiment outputs a clear image in which noise is removed and detailed information is preserved, and thus the compression efficiency is improved even when the video encoder 404 compresses the image.

Referring to FIG. 12, an image file processed by an image processing method and system according to an embodiment of the present invention is more efficient than an image file 1201 processed according to the prior art in terms of compression ratio and PSNR ratio 1202. Im knowing.

Embodiments of the invention also include computer-readable media containing program instructions for performing various computer-implemented operations. The computer readable medium may include program instructions, data files, data structures, etc. alone or in combination. The medium or program instructions may be those specially designed and constructed for the purposes of the present invention, or they may be of the kind well-known and available to those having skill in the computer software arts. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tape, optical media such as CD-ROMs, DVDs, and magnetic disks, such as floppy disks. Magneto-optical media, and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, and the like. The medium may be a transmission medium such as an optical or metal wire, a waveguide, or the like including a carrier wave for transmitting a signal specifying a program command, a data structure, or the like. Examples of program instructions include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

13 is an internal block diagram of a general purpose computer system that may be employed to construct an image processing method and system in accordance with the present invention.

Computer system 1300 includes one or more processors 1301 connected to a main memory including random access memory (RAM) 1302 and read only memory (ROM) 1303. The processor 1301 is also called a central processing unit (CPU). As is well known in the art, the ROM 1303 serves to transfer data and instructions to the CPU unidirectionally, and the RAM 1302 typically transfers data and instructions bidirectionally. Used to. RAM 1302 and ROM 1303 may include any suitable form of computer readable media. Mass storage 1304 is bidirectionally coupled to processor 1301 to provide additional data storage capability and may be any of the computer readable recording media described above. The mass storage device 1304 is used to store programs, data, and the like, and is a secondary memory device such as a hard disk which is generally slower than the main memory device. Certain mass storage devices, such as CD ROM 1306, may also be used. The processor 1301 may include one or more input / output interfaces such as video monitors, trackballs, mice, keyboards, microphones, touchscreen displays, card readers, magnetic or paper tape readers, voice or handwriting readers, joysticks, or other known computer input / output devices. 1305. Finally, the processor 1301 may be connected to a wired or wireless communication network through the network interface 1307. Through this network connection, the procedure of the method described above can be performed. The apparatus and tools described above are well known to those skilled in the computer hardware and software arts.

The hardware device described above may be configured to operate as one or more software modules to perform the operations of the present invention.

According to the present invention, there is provided a noise processing method and system capable of maintaining sharpness by preserving edge or detailed information of an image while effectively removing impulse noise or Poisson noise in the process of performing spatial filtering on the image.

As described above, the present invention has been described by way of limited embodiments and drawings, but the present invention is not limited to the above-described embodiments, which can be variously modified and modified by those skilled in the art to which the present invention pertains. Modifications are possible.

Accordingly, the spirit of the present invention should be understood only by the claims set forth below, and all equivalent or equivalent modifications thereof will belong to the scope of the present invention.

1 is a block diagram showing the configuration of a noise processing system according to an embodiment of the present invention.

2 is a flowchart illustrating a noise processing process of a noise processing system according to an embodiment of the present invention.

3 is a diagram illustrating noise processing results of a unit region including edge and detailed information and a unit region including no edge and detailed information according to an exemplary embodiment of the present invention.

4 is a diagram illustrating a case where a noise processing system according to the present invention is used as a prefilter.

5 is a block diagram illustrating an image processing system to which a noise processing system is applied according to an embodiment of the present invention.

6 is a flowchart illustrating an image processing method according to another exemplary embodiment of the present invention.

7 is a diagram exemplarily illustrating an entire image divided into a high brightness region and a low brightness region by a segmentation filter according to an embodiment of the present invention.

8 is a diagram for explaining impulse noise and Poisson noise.

9 is a diagram for explaining a misdetection motion pixel.

FIG. 10 schematically illustrates a process of removing impulse noise, Poisson noise, and false color noise from image data included in one or more segments divided into low brightness regions according to an embodiment of the present invention. Figure is shown.

11 is a view comparing the image processing result of the image processing method and system according to the present invention with the image processing result of the prior art.

12 is a view comparing the relationship between the compression ratio of the image file and the PSNR (Peak Signal to Noise Ratio) of the image processing method and system according to an embodiment of the present invention with the image processing result according to the related art. .

13 is an internal block diagram of a general purpose computer system that may be employed to construct an image processing method and system in accordance with the present invention.

<Explanation of symbols for the main parts of the drawings>

101: threshold calculation unit

102: first filter unit

103: second filter unit

Claims (5)

In the noise processing method for removing noise in the image, Dividing image data included in the image into a predetermined unit area; Calculating thresholds corresponding to the unit region by using pixel values of pixels included in the unit region; Detecting whether a first pixel including an impulse noise exists in the unit region by using the calculated threshold value; Applying a median filter to the first pixel when the first pixel is detected; And When a median filter is applied to the first pixel, identifying a second pixel adjacent to the first pixel; And Applying a mean-variance filter to the second pixel Noise processing method comprising a. The method of claim 1, Computing a threshold value corresponding to the unit region by using the pixel values of the pixels included in the unit region, The threshold Calculating with Including, remind Is a minimum threshold for detecting a pixel containing impulse noise, Is a mean value of pixel values of the pixels included in the unit region. The method of claim 1, The pixel value of the second pixel filtered by applying the mean-dispersion filter to the second pixel is Is calculated as K (i, j) denotes a first pixel value at coordinates (i, j); Is the variance of the pixel values of the second pixels, Is an average of pixel values of the second pixels. A computer-readable recording medium having recorded thereon a program for implementing the method according to any one of claims 1 to 3 on a computer. In a noise processing system for removing noise in an image, A threshold calculator for dividing the image data included in the image into a predetermined unit area and calculating thresholds corresponding to the unit area by using pixel values of pixels included in the unit area; A first filter unit detecting whether a first pixel including an impulse noise exists in the unit region by using the calculated threshold value, and applying a median filter to the first pixel when the first pixel is detected; And A second filter unit for identifying a second pixel adjacent to the first pixel and applying an average-dispersion filter to the second pixel Noise processing system comprising a.